Precision Behavioral Interventions: A Multimodal Big Data and Machine Learning Pipeline for Personalizing Early-Intervention Therapies in Children with Autism Spectrum Disorder

• Abiodun Okunola

  Ladoke Akintola University Technology

  Author

Autism Spectrum Disorder (ASD) presents a significant public health challenge, with rising prevalence and substantial heterogeneity in presentation and treatment response. Current early-intervention approaches, while evidence-based, largely follow one-size-fits-all protocols that fail to account for individual neurobehavioral variability, leading to suboptimal outcomes and inefficient resource allocation. This study addresses this critical gap by developing and validating a multimodal machine learning pipeline for precision behavioral intervention personalization in children with ASD. Leveraging a retrospective dataset of 1,247 children with ASD who received Applied Behavior Analysis (ABA) therapy over 24 months, we integrated behavioral observation data, speech and language features, physiological signals, and demographic information into a unified predictive framework. The pipeline employs a hybrid deep learning architecture combining Convolutional Neural Networks (CNN) for feature extraction from behavioral streams, Long Short-Term Memory (LSTM) networks for temporal pattern recognition, and a Deep Deterministic Policy Gradient (DDPG) reinforcement learning module for adaptive intervention optimization. Our framework achieved 89.4% accuracy in predicting individualized treatment response trajectories, significantly outperforming traditional static intervention assignment methods (p < 0.001). The reinforcement learning component demonstrated a 24.7% improvement in simulated developmental outcomes over baseline approaches. The proposed pipeline offers a replicable, data-driven framework for transitioning ASD intervention from standardized protocols to truly personalized, dynamically adaptive care, with implications for clinical practice, healthcare policy, and future precision medicine research.

• pdf

Copyright (c) 2026 Abiodun Okunola (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.